A Geometric Review of Linear Algebra

The following is a compact review of the primary concepts of linear algebra. I assume the
reader is familiar with basic (i.e., high school) algebra and trigonometry. The order of pre-
sentation is unconventional, with emphasis on geometric intuition rather than mathematical
formalism. For more thorough coverage, I recommend Linear Algebra and Its Applications by
Gilbert Strang, Academic Press, 1980.

Vectors (Finite-Dimensional)

A vector is an ordered collection of numbers, known as the components of the vector. I’ll use
the variable N throughout to represent the number of components, known as the dimension-
ality of the vector. When notating a vector, we typically arrange the components vertically (a
“column” vector):
v1
 
 v2 
~v =  ..  ,
 
 . 
vN

v
v3
Vectors of dimension 2 or 3 can be graph- v2 v
ically depicted as arrows, with the tail at
the origin and the head at the coordinate lo- v2
cation specified by the vector components. v1 v1
Vectors of higher dimension can be illus-
trated using a “spike plot”.

v1 v2 vN

pP
The norm (or magnitude) of a vector is defined as: ||~v || = 2
n vn . Geometrically, this corre-
sponds to the length of the vector. A vector containing all zero components has zero norm,
• Author: Eero Simoncelli, Center for Neural Science, and Courant Institute of Mathematical Sciences.
• Created: 20 January 1999. Last revised: 15 Sep 2010.
• Send corrections or comments to eero.simoncelli@nyu.edu
and is called the zero vector.
A unit vector is a vector of length one. In two
dimensions, a unit vector may be parameter-
ized in terms of its angle relative to the hori-
zontal axis:

cos(θ)
 v sin θ u^
û(θ) = .
sin(θ)

Angular parameterization of unit vectors v θ

can be generalized to N dimensions, but is cos θ
much messier.

2v
Multiplying a vector by a scalar simply
changes the length of the vector by that fac-
tor. That is: ||a~v || = a||~v ||. Multiplying by a
v
negative number reverses the direction. The
set of vectors corresponding to all rescal- v
ings of any given vector lie on (and cover)
a straight line. Any vector (with the excep-
tion of the zero vector) may be rescaled to
have unit length by dividing by its norm:
v̂ = ~v /||~v ||.

The sum of two vectors of the same dimen-

sionality is a vector whose components are
v translated
sums of the corresponding components of
the summands. Specifically, ~y = w
~ +~v means w
that:
w +v
z n = wn + vn , for every n

Geometrically, this corresponds to stacking

the vectors head-to-foot. v

The inner product (also called the “dot product”) of two vectors is the sum of the pairwise
product of components: X
~v · w
~≡ v n wn .
n

2
Note that the result is a scalar.

This operation has an equivalent geometric w

definition (general proof a bit tricky):

~v · w
~ ≡ ||~v || ||w||
~ cos (φvw ), φvw
where φvw is the angle between the two vec- v
tors. Thus, the inner product of two perpen- b
dicular vectors is 0, the inner product of two b
parallel vectors is the product of their norms, w . v = w .. w . v
and the inner product of a vector with itself
is the square of its norm.

The inner product is distributive over addition: ~v · (w

~ + ~y ) = ~v · w
~ + ~v · ~y . The commutative
property of scalar multiplication implies that inner products are also commutative (i.e., order
doesn’t matter): ~v · w
~ =w
~ · ~v .
Despite this symmetry, it is often useful to interpret one of the vectors in an inner product as
an operator, and the other as an input. For example, the inner product of any vector ~v with the
vector:
~ = ( N1 N1 N1 · · · N1 )
w
gives the average of the components of ~v . The inner product of vector ~v with the vector

w
~ = (1 0 0 ··· 0)

is the first component, v1 .

Furthermore, the inner product of a vector ~v

with a unit vector û has a useful geometric
v
interpretation. The cosine equation implies
that the inner product is the length of the
component of ~v lying along the line in the di- (v . u^) u^
rection of û. This component, which is writ- u^
ten as (~v · û)û, is referred to as the projection
of the vector onto the line. The difference (or
residual) vector, ~v − (~v · û)û, is the compo-
nent of ~v perpendicular to the line. Note that
the residual vector is always perpendicular
to the projection vector, and that their sum is
~v [prove].

3
Vector Spaces

Vectors live in vector spaces. Formally, a vector space is just a collection of vectors that is
closed under linear combination. That is, if the two vectors {~v , w}
~ are in the space, then the
vector a~v + bw
~ (with a and b any scalars) must also be in the space. All vector spaces include
the zero vector (since multiplying any vector by the scalar zero gives you the zero vector). A
subspace is a vector space lying within another vector space (think of a plane, slicing through
the 3D world that we inhabit).
This definition is somewhat abstract, as it implies that we construct vector spaces by starting
with a few vectors and “filling out” the rest of the space through linear combination. But we
I N)
have been assuming an implicit vector space all along: the space of all N -vectors (denoted R

4
is clearly a vector space [verify].

Working backwards, a set of vectors is said

to span a vector space if one can write any
vector in the vector space as a linear com-
v1 v3
bination of the set. A spanning set can be
redundant: For example, if two of the vec-
tors are identical, or are scaled copies of each
other. This redundancy is formalized by
defining linear independence. A set of vec-
tors {~v1 , ~v2 , . . . ~vM } is linearly independent if v2
(and only if) the only solution to the equation
X
αn~vn = 0
n

is αn = 0 (for all n).

A basis for a vector space is a linearly in-

dependent spanning set. For example, con-
sider the plane of this page. One vector is
not enough to span the plane: Scalar multi-
ples of this vector will trace out a line (which v
v
v v2
is a subspace), but cannot “get off the line” vv1 2
v1 v
to cover the rest of the plane. But two vec-
tors are sufficient to span the entire plane. v1 v
Bases are not unique: any two vectors will
do, as long as they don’t lie along the same
line. Three vectors are redundant: one can
always be written as a linear combination of
the other two. In general, the vector space
RN requires a basis of size N .

e
Geometrically, the basis vectors define a set
of coordinate axes for the space (although ê2
they need not be perpendicular). The stan- e
dard basis is the set of unit vectors that lie
along the axes of the space:

1 0 0
     
0 1 0
ê1
     
ê1 =  0 , ê2 =  0 , . . . êN =  0 .
     
 ..   ..   .. 
. . .
S x)
0 0 1 v
5 xSe
v S
xe x
xS e
Linear Systems & Matrices

A linear system S transforms vectors in one vector space into those of another vector space,
in such a way that it obeys the principle of superposition:

S{a~v + bw}
~ = aS{~v } + bS{w}.
~

x b
That is, the system “response” to any linear
combination of vectors is equal to that same + S
linear combination of the response to each of y a
the vectors alone. Linear systems are use-
ful because they are very well understood (in
particular, there are powerful tools for char- x S b
acterizing, analyzing and designing them),
and because they provide a reasonable de- +
scription of many physical systems.
y S a

The parallel definitions of vector space and linear system allow us to make a strong statement
about characterizing linear systems. First, write an arbitrary input vector ~v in terms of the
standard basis:
~v = v1 ê1 + v2 ê2 + . . . + vn êN
Using the linearity of system S, we write:

S{~v } = S{v1 ê1 + v2 ê2 + . . . + vn êN }

= v1 S{ê1 } + v2 S{ê2 } + . . . + vn S{êN }.

Input

v
That is, the response is a linear combination
of the responses to each of the standard basis
vectors. Each of these responses is a vector. v1 x v1 x

Since this holds for any input vector ~v , the

v2 x v2
system is fully characterized by this set of re- L
x

sponse vectors. v3 x v3 x

v4 x v4 x

Output

We can gather the column vectors corresponding to the responses to each axis vector into a
table of numbers that we call a matrix. This matrix is a complete representation of the associ-
ated linear system: any response is just a linear combination (weighted sum) of the columns of

6
the matrix. Label the elements of the matrix Snm with n indicating the row and m the column.
The response of the linear system to an input vector ~v has components
X
wn = Snm vm
m

The summation is over the columns of the matrix. For short, we write w ~ = S~v , and refer to
this as multiplication of the matrix by the input vector. The operation is only defined when the
number of columns in the matrix matches the dimensionality of the input vector.

An alternative interpretation of the matrix

multiplication is that each component of the 4 7 6 1 66
3
output vector is an inner product of the cor- . . . . .
responding row of the matrix with the input
vector.
. . . . .
5 = .
. . . . 2 .
. . . . 7 .

Like vectors, matrices can be multiplied by a scalar and added (element by element). In ad-
dition, the transpose is another matrix with rows and columns swapped: (S T )nm = Smn . A
symmetric matrix is a square matrix that is equal to its transpose.
The sequential application of two linear systems is a linear system. We define the matrix
associated with the full system as the matrix product of the two subsystem matrices. Let A
and B be the two subsystem matrices. By following the transformation of the standard basis
as it passes through A and then B, we can get a definition for the product matrix C:
X
Cnk = Bnm Amk
m

The columns of the product matrix are just the application of matrix B to the columns of A.
Since it can be thought of as a collection of inner products, matrix multiplication is distributive
over addition. It is also associative: A(BC) = (AB)C. But be careful: It is generally not
commutative.
It’s worth mentioning two special classes of matrix. A diagonal matrix is one for which only
elements along the diagonal can be non-zero. These matrices operate on vector spaces by
stretching or compressing axes: the nth axis of the space is stretched or compressed by an
amount specified by the nth diagonal element, Snn . The product of two diagonal matrices
is diagonal. If the matrix is square, and the diagonal elements are all one, the matrix does

7
nothing. This matrix is called the identity, denoted I.

If an element of the diagonal is zero, then

the associated axis is annihilated. The set
of vectors that are annihilated by the matrix
form a vector space [prove], which is called
the row nullspace, or simply the nullspace
of the matrix.
v2 v
[ 0 0]
2 0

v1
2v1

Another implication of a zero diagonal element is that the matrix cannot “reach” the entire
output space, but only a proper subspace. This space is called the column space of the matrix,
since it is spanned by the matrix columns. The rank of a matrix is just the dimensionality of
the column space. A matrix is said to have full rank if its rank is equal to the smaller of its two
dimensions.
An orthogonal matrix is a square matrix
whose columns are pairwise orthogonal unit
vectors. Remember that the columns of a
matrix describe the response of the system to
the standard basis. Thus an orthogonal ma-
e^2
(Ο) =
trix maps the standard basis onto a new set e^1
of N orthogonal axes, which form an alter- Ο (e^1 )
Ο (e^2 )
native basis for the space. This operation is
a generalized rotation, since it corresponds to
a physical rotation of the space and possibly
e.g. Ο = [cos θ
sin θ
sin θ
cos θ ]
negation of some axes. Thus, the product of
two orthogonal matrices is also orthogonal.
Note that an orthogonal is full rank (it has no
nullspace), since a rotation cannot annihilate
any non-zero vector.

Linear Systems of Equations

The classic motivation for the study of linear algebra is the solution of sets of linear equations
such as

a11 v1 + a12 v2 + . . . + a1N vN = b1

a21 v1 + a22 v2 + . . . + a2N vN = b2
..
.
aM 1 v1 + aM 2 v2 + . . . + aM N vN = bM

8
If we put the variables vn and the constants bm into column vectors, and the constants anm into
a matrix A, these equations may be written more compactly: A~v = ~b. Furthermore, we may
now use the tools of linear algebra to determine if there is a solution for ~v .

Inverses

What if we want to invert the action of a matrix? That is: given any output vector w, ~ find
the unique vector ~v for which S~v = w. ~ If the matrix has a (nonzero) nullspace, then this is
impossible. In particular, all vectors in the nullspace are mapped to 0, so given output w ~ =0
we cannot determine from which vector in the nullspace we started. More generally, a vector
~v satisfying S~v = w
~ is not unique, since (~v + ~n) also satisfies the expression (where ~n is chosen
as any vector in the nullspace).
If a matrix has a zero nullspace, then it can be inverted. The inverse operation has an asso-
ciated matrix, denoted S −1 . Since applying the matrix and its inverse in succession restore
the original vector, the matrix product of the inverse and the matrix should be the identity:
S −1 S = I.
For example, the inverse of a square diagonal matrix is another diagonal matrix. Since the
original matrix stretches and compresses axes, its inverse must undo these actions by multi-
plying each axis by the scalar inverse of the original stretch factor. Specifically, the inverse is
a diagonal matrix with elements Snn −1 = 1/S . Note that this cannot be done for diagonal
nn
elements equal to zero.
Orthogonal matrices also have simple inverses. Because of the orthogonality of the matrix
columns, the transpose of an orthogonal matrix is its inverse: OT O = I. Since O corresponds
to a generalized rotation of the space, OT must correspond to a generalized rotation in the
opposite direction.
Finally, we’ve neglected non-square matrices. First, consider “short and fat” matrices, which
project a vector space onto a space of smaller dimensionality. These cannot be inverted. Sup-
pose the inverse matrix S −1 were to exist. Apply it to the standard basis vectors. This produces
a set of vectors back in the input space of S. The span of these vectors is the column space of
S −1 . But this space has dimensionality equal to that of the output space of S, which we as-
sumed was smaller than that of the input space of S. Thus, our inverse cannot “reach” all the
vectors in the input space of S.
A “tall and skinny” matrix embeds a vector space into one of higher dimensionality. If the
matrix is full rank, this operation is invertible.

Singular Value Decomposition

The singular value decomposition (SVD) is a standard form for representing a matrix. It is
often taught near the end of a one-semester graduate course on linear algebra (or not at all),
probably because the proof is fairly involved. This is a shame, because it is one of the most
fundamental and beautiful results (and extremely useful as well).

9
Basic result (stated without proof): any matrix M may be decomposed into a product of three
matrices:
M = U SV T
such that U and V are orthogonal and S is diagonal with positive entries. The matrix S always
has the same dimensions as M , and the diagonal elements of S are called the singular values.
The advantage of this decomposition is that it describes the action of M in terms of easily
understood pieces: a (generalized) rotation, scaling of the axes, and another rotation. Given
this decomposition, we can directly answer many important questions regarding the matrix.
Convince yourself that the following must hold:

• the nullspace of M is spanned by those columns of V associated with zero (or non-
existent) singular values. Even better, these columns provide an orthogonal basis for the
nullspace.

• the column (target) space of M corresponds to columns of U associated with non-zero

singular values. Again, these columns provide an orthogonal basis.

• the matrix M is invertible if (and only if) the number of nonzero singular values is equal
to the number of columns of M (i.e., the dimensionality of the input space).

Note: The SVD always exists, but may be non-unique in the following ways:

1. One can permute the columns of U , diagonal elements of S, and columns of V (as long
as they’re all permuted the same way).

2. One can negate corresponding columns of of U and V .

3. Columns of U /V with equal corresponding singular values may be orthogonally trans-

formed. The same transformation must be used on columns of both U and V .

10